This is the third article from Tomasz Nurkiewicz from a series of tutorials about Akka. Things now get even more interested and you’ll learn how to change the behavior of the actors.
Interested to learn more about Scala or Akka? Then visit our Selected Books area
Sometimes our actor needs to react differently based on its internal state. Typically receiving some specific message causes the state transition which, in turns, changes the way subsequent messages should be handled. Another message restores the original state and thus – the way messages were handled before. In the previous article we implemented RandomOrgBuffer actor based on waitingForResponse flag. It unnecessarily complicated already complex message handling logic: |
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
|
var waitingForResponse = falsedef receive = { case RandomRequest => preFetchIfAlmostEmpty() if(buffer.isEmpty) { backlog += sender } else { sender ! buffer.dequeue() } case RandomOrgServerResponse(randomNumbers) => buffer ++= randomNumbers waitingForResponse = false while(!backlog.isEmpty && !buffer.isEmpty) { backlog.dequeue() ! buffer.dequeue() } preFetchIfAlmostEmpty()}private def preFetchIfAlmostEmpty() { if(buffer.size <= BatchSize / 4 && !waitingForResponse) { randomOrgClient ! FetchFromRandomOrg(BatchSize) waitingForResponse = true }} |
Wouldn’t it be simpler to have two distinct receive methods – one used when we are awaiting for external server response (waitingForResponse == true) and the other when buffer is filled sufficiently and no request to random.org was yet issued? In such circumstances become() and unbecome() methods come very handy. By default receive method is used to handle all incoming messages. However at any time we can call become(), which accept any method compliant with receive signature as an argument. Every subsequent message will be handled by this new method. Calling unbecome() restores original receive method. Knowing this technique we can refactor our solution above to the following:
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
|
def receive = { case RandomRequest => preFetchIfAlmostEmpty() handleOrQueueInBacklog() }def receiveWhenWaiting = { case RandomRequest => handleOrQueueInBacklog() case RandomOrgServerResponse(randomNumbers) => buffer ++= randomNumbers context.unbecome() while(!backlog.isEmpty && !buffer.isEmpty) { backlog.dequeue() ! buffer.dequeue() } preFetchIfAlmostEmpty()}private def handleOrQueueInBacklog() { if (buffer.isEmpty) { backlog += sender } else { sender ! buffer.dequeue() }}private def preFetchIfAlmostEmpty() { if(buffer.size <= BatchSize / 4) { randomOrgClient ! FetchFromRandomOrg(BatchSize) context become receiveWhenWaiting }} |
We extracted code responsible for handling message while we wait for random.orgresponse into a separate receiveWhenWaiting method. Notice the become() andunbecome() calls – they replaced no longer needed waitingForResponse flag. Instead we simply say: starting from next message please use this other method to handle (become slightly different actor). Later we say: OK, let’s go back to the original state and receive messages as you used to (unbecome). But the most important change is the transition from one, big method into two, much smaller a better named ones.
become() and unbecome() methods are actually much more powerful since they internally maintain a stack of receiving methods. Every call to become() (withdiscardOld = false as a second parameter) pushes current receiving method onto a stack while unbecome() pops it and restores the previous one. Thus we can usebecome() to use several receiving methods and then gradually go back through all the changes. Moreover Akka also supports finite state machine pattern, but more on that maybe in the future.
Source code for this article is available on GitHub in become-unbecome tag.
Original article can be found here : http://nurkiewicz.blogspot.gr/2012/11/becomeunbecome-discovering-akka.html
